The prevalence of AI in engineering

AI for materials classification and discovery

• MSE Professor Liz Holm uses machine vision to autonomously sort and classify materials microstructures, including 3-D printing powders. Using machine learning, Holm and her team can easily recognize whether or not a metal powder has the microstructural qualities—like strength, fatigue life, and toughness—needed for production.
• ChemE Assistant Professor Zack Ulissi uses machine learning to accelerate materials screening and catalyst discovery. Machine learning methods also allow him to predict the properties of small molecules and how they will interact with surfaces.

AI for infrastructure and buildings

• CEE Associate Professor Mario Berges is using a combination of sensors and artificial intelligence technologies to continuously monitor bridges, roadways, buildings, and other infrastructure systems in what is called indirect structural health monitoring. He also uses machine learning to study how much  energy individual appliances use, and how appliances can meet the demands of the power grid.
• CEE Professor Burcu Akinci takes a big data analytics approach to facility and infrastructure operations and management. She uses object detection and 3-D imaging to extract building information needed for data-driven decision making in managing facilities.

Powering AI systems

• ECE Professor Diana Marculescu works on making AI systems more efficient so that they can run on local devices, instead of relying solely on the cloud.
• ECE Assistant Professor Gauri Joshi designs cloud and machine learning infrastructure to reduce delays in AI systems.
• ECE Professor Franz Franchetti works on improving the hardware that supports AI systems. He works across the stack to develop AI systems that are faster, accurate, and efficient.

AI for biological applications

• BME Associate Professor Steve Chase and ECE Associate Professor Byron Yu are using machine learning to understand motor learning. They apply machine learning algorithms to brain computer interfaces to better understand and explain neural activity.
• ECE Professor Radu Marculescu uses machine learning to infer microbial relationships in humans. He developed a machine learning algorithm—called MPLasso—that mines medical and scientific literature from the past few decades in search of experimental data from research focused on various types of microbial interactions and associations. MPLasso pulls this disparate information into a centralized dataset that catalogs microbial interactions within the human GI tract.

AI in mechanical engineering

• MechE Professor Burak Kara uses machine learning to find lightweight options for 3-D printing, called topology optimization. He also developed a tool that uses AI for intuitive design. The tool recognizes what makes a car “sporty” or a shoe “comfortable.”
• MechE Associate Professor Albert Presto and his team used a machine learning approach to calibrate their low-cost air quality sensors. The calibration technique was found to improve accuracy and derive more sensitive readings.

AI and privacy

• ECE Associate Professor Lujo Bauer developed glasses that fool facial-recognition systems. He also creates machine-learning-based personal privacy assistants to help users preserve their privacy.
• ECE Professor Marios Savvides develops facial and biometric recognition systems. He works on algorithms that are accurate and more reliable when variations occur.
• ECE Professor Anupam Datta studies accountability and fairness in AI systems. He develops ways to create trust in systems, by detecting and eradicating bias.

Pervasive AI

• How can we turn chips into high-speed computers? ECE Professor Jimmy Zhu is making artificial intelligence pervasive with remanence computing. This new computer platform fuses logic, memory, and data on a single chip, enabling data-centric computing at a higher speed with fast and low power data processing.

Developing new models and methods

• EPP Assistant Professor Alex Davis worked with BME Associate Professor Adam Feinberg to develop a method to optimize 3-D printing with soft materials. The Expert-Guided Optimization (EGO) method combines expert judgment with an optimization algorithm that efficiently searches combinations of parameters relevant for 3-D printing, enabling high-fidelity soft material products to be printed.
• BME/Chemistry Professor Newell Washburn developed an algorithm that predicts and optimizes complex physical systems, using small data sets. Most machine learning algorithms require very large data sets to train the system. But in some instances researchers only have a few data points to train an algorithm.
• ChemE Professor Nick Sahinidis developed the ALAMO approach to machine learning that generates simple and accurate models for process systems engineering. While the method was developed to optimize chemical processes, it also has applications in other domains, such as thermodynamics.
• ChemE Professor John Kitchin uses machine learning to generate models for molecular simulation.
• ECE Assistant Professor Carlee Joe-Wong develops new algorithms for users in shared resource settings, such as taxi companies that compete to pick up passengers and use machine learning to learn where to go. She takes into account the user competition, and attempts to quantify the algorithm’s effectiveness when compared to non-competitive scenarios or naive ways of handling competition.